Deterministic Random Blinding for Privacy-Preserved Homomorphic Encryption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current homomorphic encryption schemes face challenges in providing privacy-preserved outsourced processing, especially when dealing with repeated operations that can reveal secret data due to statistical analysis of randomness distribution, and they require substantial computational overhead.
Innovation Solution
A deterministic but random blinding algorithm is introduced, using pseudorandom generators and cryptographic hash functions to generate consistent blinding values, ensuring the same blinded value is produced for the same secret value each time, thus preventing information leakage and reducing computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional probabilistic blinding is used to protect secret values during repeated processing, then security is maintained through randomness, but statistical analysis can reveal information about the secret value over multiple iterations
Solution Approach 1:
The system pre-generates a set of blinding values before the repeated processing operations begin. These blinding values are selected from a predetermined set rather than being generated randomly at each iteration. This preliminary preparation ensures that the same blinding value can be consistently applied across multiple operations, preventing statistical analysis from revealing information about the secret value while maintaining security.
Solution Approach 2:
The invention changes the parameter of blinding value generation from random generation at each iteration to selection from a predetermined set. This parameter change transforms the blinding mechanism from one that uses fresh randomness each time to one that uses consistent, pre-selected values, thereby eliminating the vulnerability to statistical analysis while preserving the protective function.
2Reliability
If homomorphic encryption schemes are used to enable outsourced processing of secret data, then privacy preservation is achieved, but substantial computational overhead is required
Solution Approach 1:
The invention extracts the computationally intensive random value generation step from each processing iteration. By pre-generating blinding values outside the main processing loop, the system removes the overhead of randomness generation from the critical path of repeated operations, significantly improving computational efficiency while maintaining privacy preservation through the use of these pre-generated values.
Solution Approach 2:
The system performs the computationally expensive task of generating blinding values in advance, before the actual processing operations begin. This preliminary action moves the computational burden to a setup phase rather than repeating it during each iteration, thereby improving the overall productivity and computational efficiency of the system while maintaining the privacy-preserving properties.
3Reliability
If new random blinding values are generated at each iteration for processing secret values, then security is maintained, but memory and processing resources are consumed repeatedly
Solution Approach 1:
The system recovers computational resources by reusing pre-generated blinding values across multiple iterations instead of discarding them and generating new random values each time. This recovery approach eliminates the repeated consumption of computational resources for randomness generation while maintaining security, as the same blinding values can be safely reused without compromising the protection of secret values.
Data Source
AI summary
Systems and methods include determination of a first value to be blinded, determination of a first key value, generation of a first composite value based on the first value and the first key value, performance of a hash operation on the first composite value to generate a first hash value, seeding of a pseudorandom generator with the first hash value to generate a first pseudorandom value, truncation of the first hash value based on the first pseudorandom value to generate a first truncated value, and generation of a blinded value associated with the first value based on a blinding function comprising the first value and the first truncated value.